[kaffe] make check results report

Guilhem Lavaux guilhem@kaffe.org
Thu Feb 19 04:51:02 2004


Ito Kazumitsu wrote:

>>>>>>":" == Ito Kazumitsu <kaz@maczuka.gcd.org> writes:
>>>>>>            
>>>>>>
>
>:> :> Today's "make check" showed:
>:> :> ================================
>:> :> 2 of 144 tests failed
>:> :> Please report to kaffe@kaffe.org
>:> :> ================================
>:> :> Tests that failed are:
>:> :> FAIL: TestSerialVersions.java
>:> :> FAIL: TestSerialFields.java
>:> 
>:> My last patch to libraries/javalib/java/io/ObjectInputStream.java
>:> will solve the problem, at least of TestSerialFields.java.
>
>The effect of the patch was "at least" and TestSerialVersions.java
>still fails.  But at least ArrayIndexOutOfBoundsException disappeared
>from the results of TestSerialVersions.java.
>
>By the way, with Sun's JDK, the output of TestSerialVersions.java
>looks completely different:
>
>Generating canonical stream
>  Serializing TestSerialVersions$Test0001
>    Serialization exception: java.lang.IllegalAccessException
>java.lang.NullPointerException
>	at TestSerialVersions.checkClass(TestSerialVersions.java:370)
>	at TestSerialVersions.main(TestSerialVersions.java:399)
>Exception in thread "main" 
>
>I do not know which is the right output.
>  
>
Hi !

Sorry not to have answered sooner. Both expected output and Sun's output 
are right. The only difference is that sun is checking carefully the 
authorization using java policy and kaffe is not. I've seen that 
apparently some gcj's developers broke the tests (but actually he/she 
was not aware of as it is not in mauve). I'll fix that asap.

Cheers,
Guilhem.